In that case, I should clarify what I mean by 3 losers. I mean three losers by modified losing trick count, which is a maximum of 3 losers in each suit, limited by the number of cards held in the suit, reduced by one for each of the top three honors held in the suit (top two honors in a doubleton, and K singleton is a loser).
So, for example, this hand is a 2♣ opening:
Two losers by modified losing trick count (one in hearts, one in diamonds).
But this hand is not a 2♣ opening:
Four losers by modified losing trick count (one in each suit).
Ideally, responder will be evaluating his hand by cover cards - provisionally, every A, K and Q in the hand, adjusted during the bidding as opener's distribution becomes known. Short suits can also count as cover cards with adequate trump holdings and knowledge that the shortness is working. So, opposite a 2♣ opening known to have no more than 3 losers, 2 working cover cards puts you in the slam range.
In Rosenkranz's original presentation of Romex, all opening bids and responses were based primarily on modified losing trick count and cover cards, with HCP as a secondary consideration. I don't know how that has been revised over the years, as the most recent Romex book that I have is about 30 years old and I haven't played Romex in about that long.